An AI auditor is designing a testing protocol for a loan underwriting model. Which approach BEST demonstrates that the model does not violate fair lending laws by using protected characteristics as proxies?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — you need to run disparate impact testing using legally recognized protected characteristics and apply the 4/5ths rule. That's how you show the model isn't using proxies to violate fair lending laws. Just scrubbing names, checking business purpose, or signing an attestation won't prove it.
Full Explanation
Conducting disparate impact testing using legally recognized protected characteristics and applying the 4/5ths (80%) rule is correct because it is the established regulatory method for detecting indirect discrimination in credit decisions, required under the Equal Credit Opportunity Act and aligned with CFPB guidance. Reviewing input feature names and removing protected-class references does not address proxy variables correlated with protected classes. A business justification review for commercial purposes is a legal defense, not a bias test. A signed developer fairness attestation before deployment is an administrative control, not a technical bias assessment.